An inability to accurately identify children with language impairment (LI) will impede the progress of investigations into the nature and underlying causes of language impairment. Clinically, there are monetary, time and emotional costs associated with both falsely identifying a child as LI when that child is not and in missing a child who does have a language impairment, since that child will not receive services. Early detection of LI is important to reduce the severity of the language deficit and reduce the potential negative educational and social consequences of LI. The current project is a first step in developing an alternative assessment for identifying language impairment in preschool children using a domain-referenced approach. The proposed strategy is to set up an assessment domain in such a way that each language structure included in the assessment will be shown by at least 80% of typically- developing (TD) children and no more thhe language structures. There would thus be minimal or even no overlap between the performance of the TD and LI groups and there would be a clear and significant separation between how the two groups perform. The study will include 120 participants. Participants will include children with LI, aged 3;0 to 4;0, and a group of TD children matched for age. Each child will participate in several sessions to determine eligibility for the project and language status (as TD or LI). Language samples will then be obtained during parent-child interactions. An item analysis for syntactic, semantic, and pragmatic language structures will be carried out. Structures meeting the 80%TD-50%LI criteria will be compiled into an assessment. The assessment will be piloted in two ways, 1) by recording two exemplars of each structure based on a written transcript and 2) by checking off up to two instances of each structure based on an observation of a videotaped session. Specificity and sensitivity will be ulty. The proposed development of a domain-referenced assessment has the potential to provide a tool that is more sensitive to the variable representations of language impairment in preschool children.